Demand-weighted edge betweenness centrality for identifying critical pipes in water distribution networks

Abstract

ABSTRACT Graphical abstract summarising the DWEBC workflow: a WDN is converted to a weighted graph; EPANET hydraulic outputs and demand data are used to compute DWEBC; and high-ranking pipes support resilience screening, maintenance prioritisation, and digital-twin decision-making. Water distribution networks (WDNs) support urban resilience by delivering potable water under highly variable demands and hydraulic conditions. Traditional graph-theoretic centrality measures identify critical pipes based solely on topology, overlooking the spatial heterogeneity of nodal demands that drives service loss priorities. This study proposes demand-weighted edge betweenness centrality (DWEBC), a hydraulically informed extension of EBC in which pipe length, head loss, and target-node demand are embedded into the edge-weight function, thereby aligning shortest-path computations with the hydraulic relevance of the paths. Three weighting schemes – global constant, time-varying nodal, and time-varying nodal and pipe-specific scaling – are formulated. The algorithm is demonstrated on a small laboratory-scale Kaina network, Exnet, d-town, and d-town benchmark networks and validated on Net3, where DWEBC, in particular the time-varying nodal and pipe-specific variant, distinctly ranks pipes that serve major demand corridors. This offers a rapid screening tool for maintenance prioritisation, emergency and capital planning, and resilience assessment. Computational performance outruns several approaches by orders of magnitude, enabling routine re-analysis under shifting demand scenarios. It can be easily scaled, and it can make use of real-time flow telemetry to enable dynamic and hybrid frameworks, such as Digital Twin, to enable real-time and realistic network performance analysis and control.